CROSS REFERENCE TO RELATED APPLICATIONSThis is a continuation of PCT/US02/37589, filed Nov. 22, 2002 in the English language and designating the U.S., which, in turn, is based on and derives the benefit of the filing date of U.S. Provisional Application No. 60/341,260, filed Dec. 20, 2001, the contents of each of which are incorporated herein by reference in their entireties.[0001]
BACKGROUND OF THE INVENTION1. Field of the Invention[0002]
The present invention relates to plasma processing systems of the type that may be used, for example, for deposition of material on or the etching of material from a workpiece. The invention is more specifically directed to a method and apparatus for improving the properties of a plasma used to process a workpiece.[0003]
2. Background of the Invention[0004]
A process plasma is a collection of charged particles and radicals that may be used to process (that is, to remove material from or deposit material on) a workpiece. Process plasmas are used in the manufacture of integrated circuit (IC) devices, flat panel displays and other products. Process plasmas may be used, for example, to etch (i.e., remove) material from or to sputter (i.e., deposit) material on a workpiece in the form of, for example, a semiconductor wafer during IC fabrication.[0005]
A reactive process plasma may be generated by introducing a process gas into a plasma chamber and then ionizing and dissociating the gas. Plasma generated in the chamber strikes the workpiece during processing of the workpiece. The quality and efficiency of commercial plasma processing operations can be improved by improving the characteristics of the process plasma and the methods of generating the same.[0006]
SUMMARY OF THE INVENTIONThe present invention provides methods and apparatuses for processing a workpiece with a plasma. An illustrative embodiment of the apparatus includes a source gas injection device constructed and arranged to inject a gaseous source material into a source region of the apparatus and a plasma generating device mounted in plasma generating relation to the source region. The plasma generating device is constructed and arranged to transmit energy to a gaseous source material in the source region to generate a source plasma. The apparatus includes a process gas injection device constructed and arranged to inject a gaseous process material into a process region of the apparatus and a magnetic filter assembly constructed and arranged to impose a magnetic field generally between the source region and the process region to control the flow of charged particles from the source plasma into the gaseous process material to generate a process plasma in the process region. A source electrode is in contact with the source plasma and is constructed and arranged to control the potential of the source plasma. The apparatus includes support structure to support a workpiece so that the charged particles strike the workpiece.[0007]
An example method for processing a workpiece includes generating a source plasma, providing a process gas, controlling a flow of charged particles from the source plasma into the process gas to generate a process plasma from the process gas and to control properties of the process plasma, and striking the workpiece with charged particles from the process plasma.[0008]
BRIEF DESCRIPTION OF THE DRAWINGSFIG. 1 shows a schematic view of an illustrative embodiment of a plasma processing system constructed according to the principles of the present invention;[0009]
FIG. 2 shows a schematic cross-sectional view of an illustrated embodiment of a plasma processing apparatus in isolation constructed according to the principles of the present invention;[0010]
FIG. 3 is a schematic drawing showing a plan view of a plasma potential control electrode of the apparatus of FIG. 2 in isolation;[0011]
FIG. 4 is a schematic drawing showing a cross-section of the electrode taken along the line[0012]4-4 of FIG. 3;
FIG. 5 is a schematic drawing showing a plan view of a magnetic filter assembly of the apparatus of FIG. 2 in isolation;[0013]
FIG. 6 is schematic drawing showing an elevational view of the magnetic filter assembly;[0014]
FIG. 7 is a schematic drawing showing an enlarged view of the portion of the magnetic filter assembly enclosed within the circle formed by the broken line in FIG. 6;[0015]
FIG. 8 is a schematic drawing showing a bottom plan view of a process gas injection device of the apparatus of FIG. 2 in isolation;[0016]
FIG. 9 is a schematic drawing showing a cross-section of the process gas injection device taken along the line[0017]9-9 of FIG. 8;
FIG. 10 is a schematic drawing showing a spaced arrangement of a plurality of current conductive members and indicating the direction of current flow in each member;[0018]
FIG. 11 is a schematic drawing similar to the FIG. 10 except showing an additional row of current conductive members and indicating the direction of current flow in each member;[0019]
FIG. 12 is a schematic drawing showing a spaced arrangement of a plurality of permanent magnets and indicating the direction of magnetic polarity of each permanent magnet;[0020]
FIG. 13 is a schematic drawing similar to FIG. 1 except showing another combination of current flows in the members;[0021]
FIG. 14 is a schematic drawing similar to FIG. 12 except showing another arrangement of polarities;[0022]
FIG. 15 shows another example of a magnetic filter assembly;[0023]
FIG. 16 shows another example of a magnetic filter assembly; and[0024]
FIG. 17 shows an example of a magnetic filter assembly constructed by interengaging the magnetic filter assemblies of FIGS. 15 and 16.[0025]
DETAILED DESCRIPTION OF THE EMBODIMENTSFIG. 1 shows a schematic representation of an example of a[0026]plasma processing system10 that includes aplasma processing apparatus12 constructed according to the principles of the present invention. Theplasma processing apparatus12 is shown schematically in isolated view in FIG. 2. Theplasma processing apparatus12 includes areaction chamber14 having aninterior area16. Theinterior area16 includes asource region18 for containing and supporting a source plasma20 (see FIG. 2) and aprocess region22 for containing and supporting aprocess plasma24.
A plasma generating device in the form of a spiral- or coil-shaped radio frequency (RF)[0027]antenna26 is mounted on thereaction chamber14. In the illustrative embodiment of FIGS. 1 and 2 is a transformer coupled plasma (TCP) chamber and theantenna26 is in the form of a spiral antenna orspiral member26. Thespiral member26 is constructed and arranged to transmit energy to a gaseous source material in thesource region18 to generate thesource plasma20. Thespiral member26 may be operated to generate an inductively coupled plasma (ICP) in thesource region18. Thespiral member26 may be in electrical communication with apower source28 through amatching network30. Thepower source28 is capable of transmitting an RF power signal. The matchingnetwork30 may be inserted between theRF power source28 and thespiral member26 in order to maximize the power transferred from thepower source28 to thespiral member26 and thereby maximize the power transferred from thespiral member26 to asource plasma20.
The[0028]spiral member26 may be covered by ashielding structure32 mounted on thereaction chamber14. Theshielding structure32 may be constructed of a material appropriate to provide shielding of the energized spiral member26 (such as a conductive material, for example). Theshielding structure32 may also improve the efficiency of power transfer from thepower source28 to thespiral member26. Thespiral member26 may also be mounted on asupport structure34 which may be constructed of a dielectric material to facilitate transmission of the RF power generated by the energizedspiral member26 to thesource region20 of thereaction chamber14. More specifically, the top wall of thereaction chamber14 includes an opening which is covered by a dielectric material comprising thesupport structure34. Thesupport structure34 seals the opening to allow a vacuum to be created in theinterior area16 of thereaction chamber14, but allows the RF power to enter theinterior area16. The energizedspiral member26 may be operated to transmit energy to a gaseous source material in thesource region18 to produce an inductively coupledsource plasma20 having a relatively high uniform density.
The[0029]reaction chamber14 includes anouter wall40 which may be constructed of an appropriate metal material, such as aluminum. Thereaction chamber14 includes one or more outer side wall portions that at least partially surround thesource region18 and theprocess region22. Thewall40 may be in electrical communication with a ground potential during plasma processing.
A source electrode in the form of a plasma[0030]potential controlling electrode36 is mounted within thereaction chamber14 to control the potential of thesource plasma20. The potential controllingelectrode36 is mounted on anisolator support38 which is a non-electrically conductive structure that electrically isolates theelectrode36 from thewall40 of thereaction chamber14. Theelectrode36 is in electrical communication with anRF power source42 through amatching network44. Thematching network44 may be inserted between theRF power source42 and theelectrode36 in order to maximize the power transferred to asource plasma20 by the plasmapotential controlling electrode36. Theelectrode36 includes at least one surface that is substantially in contact with thesource plasma20.
An electrode assembly in the form of a[0031]chuck electrode46 is mounted on a side of thechamber14 opposite the side of thechamber14 on which thespiral member26 is mounted. Thechuck electrode46 provides support structure within the chamber that functions to support a workpiece48 (which may be a semiconductor wafer, for example). Thechuck electrode46 may also be energized to generate a potential that attracts charged particles from theprocess plasma24 towards theworkpiece48 so that the charged particles strike theworkpiece48 to etch material from theworkpiece48 or to sputter material on theworkpiece48. Theexample chuck electrode46 is movably mounted in thereaction chamber14 for movement generally toward and away from theprocess region22 to adjust the distance between a process plasma supported in the process region of thechamber14 and theworkpiece48. More specifically, thechuck electrode46 is supported by a mechanical assembly (not shown) that is sealed within aflexible bellows structure50 so that thechuck electrode46 and theworkpiece48 are axially movable prior to or during a plasma processing operation. Thechuck electrode46 may be in electrical communication with apower source52 through amatching network54 to maximize power transfer.
The[0032]example chuck electrode46 is an electrode that is in electrical communication with anRF power source52. Thechuck electrode46 may have a ground voltage or an RF bias during a plasma processing operation. The electrical path to thechuck electrode46 may further comprise animpedance match network54 which may be used to optimize power transfer through thechuck electrode46. The electrical bias of a chuck electrode is well known to those of skill in the art.
The[0033]apparatus12 includes a sourcegas injection device56 and a processgas injection device60. The sourcegas injection device56 is coupled to a sourcegas supply system58 which operates to supply the one or more gases injected into thesource region18. The sourcegas injection device56 is mounted in thereaction chamber14 in the vicinity of thespiral member26 and is operable to inject a source gas (or gasses) into thesource region18. An example of the sourcegas injection device56 is a substantially annular structure. The sourcegas injection device56 may be mounted generally about the periphery (i.e., 360°) of thespiral member26 and may operate to inject (about the 360 degree periphery of the spiral member26) and distribute one or more source gases toward the center of thesource region18 as indicated by the directional arrows Gs in FIG. 2. The source gas material may include, for example, a carrier gas (such as argon) and/or an etch gas.
The process[0034]gas injection device60 includes an array oftubes62 which, for example, may be equally spaced from one another and mounted across thereaction chamber14 as shown, for example, in FIG. 2. The plurality oftubes62 are coupled to a processgas supply system64. Thesupply system64 may supply one or more process gases for injection into the process region of thereaction chamber14. Eachgas injection tube62 includes one or more gas outlet openings (not shown) which are oriented to direct the gaseous processing material into theprocess region22 of thereaction chamber14 as indicated by directional arrows Gp in FIG. 2. The array oftubes62 may provide a symmetric array or other symmetric or asymmetric arrangement of gas outlet openings in theexample reaction chamber14 which distribute the process gas or gases in theprocess region22. Acover structure66, which may be constructed of silicon, may be mounted adjacent the array oftubes62 and between thetubes62 and thechuck electrode46 to protect the gas distribution tubes from plasma bombardment.
A selected gas (or gasses) may be supplied to the source[0035]gas injection device56 and/or the processgas injection device60 to purge thechamber14, for example, or to serve as a source gas or process gas, respectively, for plasma formation in thechamber interior16. Theplasma processing apparatus12 includes avacuum system72 coupled to theplasma chamber14 through a vacuum line. Thevacuum system72 may be coupled to thereaction chamber14 through a gas outlet opening74 as shown in the schematic view of FIG. 2 for removal of gases from theinterior area16 of thereaction chamber14.
A[0036]magnetic filter assembly68 is mounted within thereaction chamber14 generally between the plasmapotential controlling electrode36 and the processgas injection device60. As considered in detail below, themagnetic filter assembly68 imposes a magnetic field generally between thesource region18 and theprocess region22 to control the flow of charged particles from asource plasma20 in thesource region18 into the process gas to generate aprocess plasma24 in theprocess region22. Themagnetic field assembly68 may include permanent magnets, electromagnets or a combination of both. The magnetic filter assembly is considered in detail below. Themagnetic filter assembly68 is in electrical communication with apower source70 which may provide themagnetic filter assembly68 with either a DC current or an RF current.
The[0037]spiral member26 and theelectrodes36,46 may be independently cooled by a fluid that circulates from acooling system76, through one or more fluid chambers (not shown) associated with thespiral member26 and/or with eachelectrode36,46, and then back to thecooling system76.
The[0038]plasma processing apparatus12 may optionally include a plurality of voltage probes (not shown) in the form of a plurality of electrodes. Each electrode may be capacitively coupled to a respective transmission line between anRF power source28,42,52, or70 and the associateddevice26,36,46, or68. An example voltage probe is described in detail in commonly assigned pendingU.S. patent application 60/259,862 (filed on Jan. 8, 2001), which application is incorporated in its entirety herein by reference. Theplasma processing apparatus12 may optionally include anoptical probe78 for determining plasma characteristics and conditions based on spectral and/or optical properties of the plasma.
The[0039]example apparatus12 also includes acontrol system80 which is electrically communicated to various components of theapparatus12 to monitor and/or control the same. Thecontrol system80 is in electrical communication with and may be programmed to control the operation of thegas supply systems58,64, thevacuum system72, thecooling system76, the voltage probe (not shown), theoptical probe78, and eachRF power source28,42,52,70. The matching networks30,44,54 may optionally be coupled to and controlled by thecontrol system80. Moreover, the electro-mechanically operatedtranslation stage50 forchuck electrode46 can be operated and controlled via commands fromcontrol system80.
The[0040]control system80 may send control signals to and receive input signals (feedback signals, for example) from thesystem components58,64,72,76,78,28,42,52,70,30,44,54,50 and the voltage probes. Thecontrol system80 may monitor and control the plasma processing of a workpiece. As will become apparent, in an instance in which themagnetic filter assembly68 includes one or more electromagnets, thecontrol system80 may be programmed to control the power source powering each electromagnet and thereby control the passage of or the “filtering” of charged particles, particularly electrons, from the source plasma into the process region. It can be understood that although FIG. 1 shows asingle power source70 in electrical communication with themagnetic filter assembly68, in the instance in which themagnetic filter assembly68 includes more than one electromagnet, theapparatus12 may include an equal number of power sources so that, in some embodiments, each electromagnet of the magnetic filter assembly may be in electrical communication with a respective independently controllable power source.
The[0041]control system80 may be provided by a computer system that includes a processor, computer memory accessible by the processor (where the memory is suitable for storing instructions and data and may include, for example, primary memory such as random access memory and secondary memory such as a disk drive) and data input and output capability for communication of data to and from the processor.
The methods of the present invention can be illustrated with reference to the example[0042]plasma processing system10. The operation of theplasma processing system10 can be understood with reference to FIG. 1. A workpiece (or substrate)48 to be processed is placed on a support surface provided by thechuck electrode46. Thecontrol system80 activates thevacuum system72 which initially lowers the pressure in theinterior area16 of theplasma chamber14 to a base pressure (typically 10−7to 10−4Torr) to assure vacuum integrity and cleanliness of thechamber14. Thecontrol system80 then raises the chamber pressure to a level suitable for forming a source plasma and for processing theworkpiece48 with the plasma (a suitable interior pressure may be, for example, in the range of from about 1 mTorr to about 1000 mTorr). In order to establish a suitable pressure in thechamber interior16, thecontrol system80 activates thegas supply system58 and/or the processgas source system64 to supply a source gas and/or a process gas through thegas injection devices56,60, respectively, to thechamber interior16 at a prescribed flow rate (or rates) and thevacuum system72 may be throttled, if necessary, using a gate valve (not shown).
The[0043]control system80 then activates theRF power sources28,42,52,70 to power thespiral member26, the plasmavoltage controlling electrode36, thechuck electrode46 and themagnetic filter assembly68. TheRF power sources28,42,52,70 may provide voltages to thespiral member26, theelectrode36, thechuck electrode46, and themagnetic filter assembly68 at selected frequencies (including a zero frequency in the instance in which a ground or other constant voltage is applied). Thecontrol system80 may, during a plasma processing operation, independently control theRF power sources28,42,52,70 to adjust, for example, the frequency and/or amplitude of the voltage transmitted by each power source.
The[0044]RF power source28 may be operated to energize thespiral member26 to convert the source gas to a source plasma. The ICP in thesource region18 has a short, flat volume and a relatively high density. The plasma density of the source plasma may be, for example, in the range of from about 1×1012cm−3to about 5×1012cm−3. A typical electron temperature is, for example, several electron volts and a typical operating pressure is, for example, in the 10's of mTorr range. The plasma potential of the source plasma will be, for example, in the 10's of volts range. The operator or thecontroller80 can control the potential of the source plasma by controlling the DC or RF bias voltage applied to thepotential controlling electrode36. When a plurality of electrodes are in contact with the plasma, each electrode of the plurality having a comparable amount of surface area in contact with the plasma, the source plasma takes on the potential of the most positive electrode. In the instance in which thepotential controlling electrode36 is grounded, the source plasma may only assume a potential slightly higher than ground potential. In the steady state, the total electron current leaving the source plasma may be balanced by the total ion current to thesystem wall40. When thepotential controlling electrode36 is biased to a voltage significantly above ground potential, the electron current attracted to thepotential controlling electrode36 increases. When this occurs, the total electronic current lost from the source plasma may exceed the total ion current, and the plasma potential may be forced to increase until a new steady state is reached at which the electron and ion currents are about equal to one another. In this way, a DC bias on thepotential controlling electrode36 can control the plasma potential of the source plasma. If the ground surface area is large, increasing ion current to the grounded surface may not only increase the ion energy loss, but it may also increase the ion sputtering from the grounded surface. Therefore, RF bias instead of DC bias may be used for controlling the potential of the source plasma. It has been commonly known by those skilled in the art of RF plasmas that a RF bias may cause a self-generated DC bias on the electrode that is positive with respect to the ground potential if the ground surface area is comparable to the surface area of the electrode without causing additional sputtering.
The construction of the example[0045]potential controlling electrode36 can be understood from the FIGS. 3 and 4. The plasmapotential controlling electrode36 is comprised of a plurality ofelongated structures82 which are secured within anannular frame84. Theelongated structures82 are essentially parallel to one another and are spaced apart to provide a series of slot-like openings of various sizes and shapes, generally designated86. Theelongated structures82 and theframe84 are each constructed of an electrically conductive material and may be constructed of aluminum, for example, or other suitable metal material. Theopenings86 are large enough to provide a high degree of plasma flow transparency from thesource region18 to theprocess region22. The total surface area of theelectrode36 is large enough so that at least a portion of thesource plasma20 is in substantial contact with theelectrode36 during processing. Thesource plasma20 assumes the potential of the surface having the highest potential that is substantially in contact with thesource plasma20. Theelectrode36 is constructed so that it provides a surface or surfaces that are in substantial contact withsource plasma20. The surface area of thepotential controlling electrode36 that is in contact with thesource plasma20 is large relative to the surface area of the remaining surfaces (includingwalls40, for example) surrounding thesource plasma20. In this way, the voltage ofsource plasma20 can be controlled by controlling the voltage of theelectrode36.
The[0046]source plasma20 diffuses through theopenings86 in theelectrode36, through slots oropenings88 in themagnetic filter assembly68, through slots or openings90 in the processgas injection device60 and into theprocess region22 and forms aprocess plasma24 there. The structure of themagnetic filter assembly68 can be understood from FIGS. 5-7.
The[0047]magnetic filter assembly68 is an illustrative embodiment comprised of a plurality of bar-typepermanent magnets92, each of which is mounted within atubular housing94. Eachbar magnet92 extends the length of the associatedtubular housing94 so that eachmagnet92 extends between theside walls40 of the chamber when themagnetic filter assembly68 is mounted in theapparatus12. Eachhousing94 may be constructed of an electrically conductive material such as an appropriate metal material or, alternatively, eachhousing94 may be constructed of a dielectric material, as another alternative, eachhousing94 may be made of both a metal material and a dielectric material. The example housings94 are constructed of aluminum. Theassembly68 further includes aframe96 having acentral opening98. Eachhousing94 is secured within a pair ofopenings100 on opposite sides of theframe96. A pair ofannular wall structures102,104 are mounted about theopening98 on respective opposite sides thereof. Theannular wall structures102,104 may form part of thewall40 of thereaction chamber14 when theframe96 is mounted in theapparatus12. Thehousings94 of themagnetic filter assembly68 may be either in electrical communication with a ground potential or in electrical communication with a floating (for example, RF) potential. As mentioned, thehousings94 are spaced to define a plurality ofopenings88 therebetween.
The details of the construction of the process[0048]gas injection device60 can be understood from FIGS. 8 and 9. Eachtube62 of the processgas injection mechanism60 may have a generally rectangular transverse cross-section and may be constructed of an appropriate metal material such as aluminum. Thetubes62 are secured within an annulartubular frame106 which may be constructed of an appropriate metal material such as aluminum. Thetubes62 may be secured to theframe106 by welding or other appropriate method such that gas that is introduced into theframe106 flows into and through thetubes62. The process gas supply system64 (shown schematically in FIG. 8) may be coupled todevice60 through agas inlet107 in theframe106. Thetubes62 have a series ofgas outlets109 through which the gas flows into the apparatus. As mentioned, thetubes62 are spaced to provide a series of openings90 therebetween. The processgas injection device60 may optionally be in electrical communication with a grounded potential or to a floating potential.
As best seen in FIG. 2, the[0049]openings86,88,90 in theelectrode36, thefilter assembly68, and the processgas injection device60, respectively, are aligned with one another (vertically aligned in the example apparatus12) and each series of vertically alignedopenings86,88,90 in thecomponents36,68,60 are of approximately equal dimensions to one another to facilitate passage of charged particles from thesource region18 to theprocess region22.
Electrons and charged particles (example, positive ions) from the source plasma in the source region diffuse through the[0050]openings86,88,90 into theprocess region22 and form aprocess plasma24 there. Themagnetic filter assembly68 imposes a magnetic field between thesource region18 and theprocess region22 which tends to filter out the high energy electrons and prevent them from diffusing into theprocess plasma24. Energetic electrons whose mean free path of collision is longer than the magnetic field scaling (that is, the size of the magnetic field region) are reflected by the magnetic field across the openings90 in themagnetic filter assembly68 and are thereby prevented from entering theprocess plasma24. Therefore, the electrons in theprocess plasma24 have a lower average energy (that is, a lower electron temperature) than the electrons in thesource plasma20. A typical electron temperature of aprocess plasma24 is one electron volt or less, depending on the magnitude of the magnetic fields imposed by themagnetic filter assembly68. Generally, the stronger the magnetic field imposed by themagnetic field assembly68, the lower the electron temperature and the lower the electron density of theprocess plasma24.
The ions moving from the[0051]source plasma20 to theprocess plasma24 will be accelerated (downward in the example apparatus12) with an energy approximately equal to the difference between the source plasma potential (VSP) and the process plasma potential (VPP) times the ionic charge. The ion energy can be expressed by the following formula:
Ei=q(VSP−VPP), (Equation 1)
where E[0052]iis the ion energy, q is the charge on the ion, VSPis the plasma potential of thesource plasma20 and VPPis the plasma potential of theprocess plasma24. In the instance in which the potential of theprocess plasma24 is maintained at ground potential, the ion energy is determined by the bias voltage applied to thepotential controlling electrode36. This relation to can be expressed mathematically as shown in Equation 2:
Ei≈q(VSP). (Equation 2)
When the[0053]housings94 of themagnetic filter assembly68 have a ground potential, the ions coming from thesource plasma20 will be directed toward theprocess plasma24 with an energy comparable to the difference between the potential of thesource plasma20 and the potential of theprocess plasma24 times the ionic charge (as given by Equation 2). The ions entering theprocess plasma24 may thus attain an energy that is directed toward theworkpiece48 before they enter the process plasma sheath. Etching is more effective when carried out with ions having an energy directed toward theworkpiece48. When the direct energy of the ions is high enough, a high etch rate can be achieved while the workpiece48 (e.g., a semiconductor wafer) is grounded.
It can thus be understood that the magnetic field imposed by the[0054]magnetic filter assembly68 can function to separate the plasma volume within theinterior area16 of thereaction chamber14 into twoplasmas20,24 occupying two regions: the highdensity source plasma20 in thesource region18 in which the primary electrons are highly effectively confined and the relativelycool process plasma24 in theprocess region22 having very few or no ionizing electrons.
In DC plasma and in RF plasmas, the primary electrons that are accelerated directly by the external electric field are responsible for production of the[0055]process plasma24 and for the shape of the electron energy distribution function (EEDF) of theprocess plasma24. The primary electrons with energy of between twenty and one hundred electron volts (20 to 100 eV's) generally have a mean free-path of ionization collisions that is much longer than any of the dimensions of thesource region18 of thereaction chamber14 containing thesource plasma20. Confining the primary electrons to thesource plasma20 in thesource region18 increases the density of theprocess plasma24, which in turn increases the production efficiency of thesystem10, and also improves the uniformity of theprocess plasma24 and of the processedworkpiece48.
The EEDF of a[0056]process plasma24 can be controlled by controlling the strength of the magnetic field imposed by themagnetic filter assembly68. The magnetic field of themagnetic filter assembly68 can be controlled by providing individual magnets in the array of different strengths, by controlling the spacing and number of magnets, and so on. The controlling of the strength of the magnetic field controls the dissociation of the process gases because the dissociation process which occurs in theprocess plasma24 will depend to a large extent on the electron energy in the process plasma. Because the imposed magnetic field functions to reflect high energy electrons and thereby confine them to thesource region18, the process plasma has a relatively low electron temperature. Therefore, adjustment of the magnetic field strength of the magnetic field imposed bymagnetic filter assembly68 can lead to control of the plasma chemistry ofprocess plasma24.
It has been found experimentally that a magnetic field having a magnetic flux of 200-300 G-cm is strong enough to separate a highly[0057]ionized source plasma20 having an electron temperature of from approximately 4 to approximately 5 eV from a relativelycool process plasma24 having an electron temperature of less than 1 eV. Control of the electron energy of theprocess plasma24 in the range of from approximately 1 to approximately 3 eV can be achieved by varying the magnetic flux imposed by themagnetic filter assembly68 in the range of from 100 to 200 G-cm. Examples of magnetic filter assemblies that include electromagnets which provide variable magnetic fields which covers these example limits and a wider range of limits are described below.
Many different embodiments of the magnetic filter assembly are contemplated and within the scope of the invention. For example, the magnetic filter assembly could include an array of[0058]permanent magnets92 as shown in FIGS. 5-7. When an array of permanent magnets92 (represented schematically by rectangles in FIGS. 12 and 14) are used to create the imposed magnetic field, the poles of the magnets can be arranged in several different ways. For example, the array ofmagnets92 could be arranged so that all of the north poles “face” in the same transverse direction. In the illustrative embodiment of FIGS. 5-7, for example, thepermanent magnets92 that extend the lengths of the associatedtubular housings94 could be arrangement such that the north poles thereof face in the same direction. This type of arrangement can be understood from the schematic diagram of FIG. 14. FIG. 14 shows fourmagnets110,112,114,116 and thespaces118 therebetween. The magnets are arranged so that the north pole of each magnet is directed toward the one side of the chamber and the south pole of each magnet is directed toward the opposite side of the chamber. Another example arrangement that can be used when the magnetic filter assembly is comprised of a series of parallel bar-type magnets is shown schematically in FIG. 12. In this example, themagnets110,112,114,116 are arranged such that the north pole of every other magnet (110 and114, for example) faces in one direction and such that the north pole of each magnet therebetween (112 and116, in this example) faces in the opposite direction. When greater (or lesser) numbers of permanent bar magnets are used to construct the magnetic fields assembly, these patterns are simply repeated. It can be understood that these arrangements of the poles are examples only and that many other possible combinations are possible.
The magnetic fields of a magnetic filter assembly can also be provided by electromagnets. For example, a magnetic filter assembly could be comprised of an array of current-carrying members that are generally parallel to one another, that are spaced apart from one another, and that extend generally transversely across the[0059]interior area16 of thereaction chamber14 in a manner similar to the arrangement and spacing of the permanent magnets shown in FIG. 2, for example. More specifically, in one embodiment of this type of magnetic filter assembly, one or more current carrying members could be mounted within and extend the length of a respective tubular housing94 (in place of the permanent magnet therein) of the magnetic filter assembly of FIG. 5, for example. Each current carrying member may be in electrical communication with a source of current (not shown). FIG. 10 shows an example of an array of current-carrying members (which could be, for example, an array of rigid structures each of which is constructed of an electrically conductive material such as a suitable metal material and each of which could be in electrical communication with a respective controllable current source) which could provide the magnetic fields for a particular magnetic filter assembly.
Specifically, FIG. 10 shows four example current-carrying[0060]members120,122,124,126 and thespaces128 therebetween. Charged particles moving from a source plasma into a process plasma would pass through thespaces128. Each elongated current-carrying member is shown schematically in end view as a circle. The circles having “dotted” centers (members120,124) represent current-carrying members in which the current is flowing toward the viewer (that is, “out of” the page). The circles having an “X” in the center (members122,126) represent current-carrying members in which the current is flowing away from the viewer (that is, “into” the page). The arrangement of current-carrying members and the distribution of currents flowing therein in FIG. 10 is generally referred to as a “single picket fence”. A single picket fence creates a current grid that extends across the chamber. Eachmember120,122,124,126 may comprise a separate DC electromagnet.
FIG. 11 shows another example arrangement of current-carrying members. This arrangement includes a second layer of current-carrying[0061]members130,132,134,136. In this arrangement, each current carryingmember130,132,134,136 may be enclosed within a tubular housing in a manner similar to the manner in which thepermanent magnet92 is mounted in thehousing94 of FIG. 5, for example. That is, a current carrying member may extend through the associated housing94 (in place of the permanent magnet92) and be in electrical communication with a controllable source of current. The current-carryingmembers130,132,134,136 are generally parallel to one another and are generally parallel to and vertically spaced from the first layer of current-carryingmembers120,122,124,126. The arrangement of FIG. 11 is generally referred to as a “double picket fence”. The double picket fence uses two current layers (current-carryingmembers120,122,124,126 providing the first layer and current-carryingmembers130,132,134,136, providing the second layer) to impose a filtering magnetic field between the source region and the process region.
Many current flow patterns can be achieved with this double picket fence structure. In the example shown in FIG. 11, the directions in which the current is flowing in each current-carrying member is indicated with dots and X's in the manner described above. The current patterns shown in FIG. 11 create a magnetic field similar to the one created by the array of permanent magnets in FIG. 12.[0062]
FIG. 13 shows an arrangement of current-carrying members identical to that shown in FIG. 11. The current flows that are illustrated in FIG. 13 create, in effect, two current “sheathes”, the flow of current in one sheath (comprised of current-carrying[0063]members120,122,124,126) going in one direction, and the current in the other sheath (comprised of current-carryingmembers130,132,134,136) going in the opposite direction. This arrangement of current-carrying members and current flow distributions creates a field generally referred to as a “magnetic wall”. In a magnetic wall, the magnetic field is going in one direction. The arrangement of current-carrying members and the currents flowing therein shown in FIG. 13 create a magnetic field similar to the magnetic field created by the array of permanent magnets shown in FIG. 14.
Each example arrangement of FIGS. 10-14 produces a different magnetic field. One advantage of using currents to create magnetic fields (as in FIGS. 10, 11 and[0064]13) is that the magnetic field strength can be varied. The magnitude of the currents required to create magnetic fields of sufficient strength to filter electrons effectively in thereaction chamber14 can be quite high, however, and may be a few hundred amperes, for example. Arrangements such as those shown in FIGS. 12 and 14 are advantageous because they do not require an external power supply, but the magnetic fields created by a fixed array of permanent magnets cannot be varied.
FIGS. 15-17 show examples of three[0065]magnetic field assemblies138,140,142 which provide variable magnetic fields without requiring high currents. FIG. 15 shows aU-shaped member139 havinglegs144 which extend from abight portion146. TheU-shaped member139 comprises a horseshoe-type electromagnet. An array ofpermanent magnets148 are each connected to thebight portion146 of theU-shaped member139 and extend outwardly therefrom in spaced relation to theleg portions144. TheU-shaped member139 can be energized to create an electromagnet. The U-shaped member may, for example, be an integral structure that is constructed of a magnetic flux-conducting material (such as iron or other suitable metal material) that is shaped to define the bight andleg portions146,144. One ormore coil magnets150 are wound around thebight portion146 of theU-shaped member139. Eachcoil magnet150 may be in electrical communication with a respective controllable current source (not shown). When eachcoil magnet150 is energized, magnetic field energy generated by thecoil magnets150 is conducted through theU-shaped member139. TheU-shaped member139 may comprise a horseshoe-type magnet. Thepermanent magnets148 improve the properties of thehorseshoe magnet139. More specifically, thepermanent magnets148 contribute some DC magnetic field to make the magnetic field generated by theU-shaped member139 when thecoil magnets150 are energized to increase the total field strength generated by theassembly138. Thepermanent magnets148 also tend to concentrate the magnetic flux produced by theU-shaped member139/coil magnets150 in thecenter region151 of themagnetic field assembly138.
The[0066]magnetic filter assembly140 shown in FIG. 16 is similar in construction to themagnetic filter assembly138 of FIG. 15, except that thebight portion152 of theassembly140 is longer than thebight portion146 ofassembly138 and themagnetic filter assembly140 includes threepermanent magnets154 connected to thebyte portion152 thereof. A plurality ofconductive coils153 are wound around thebight portion146 of theassembly140.
Each[0067]magnetic filter assembly138 and140 may be contained within a housing. FIG. 15 shows an example of a housing153 (in dashed lines) around themagnetic filter assembly138. Thehousing153 may be constructed of a metal material such as aluminum or other appropriate material. Thehousing153 includes atubular body portion155 and a plurality oftubular arms157 which extend outwardly from thebody portion155 and surround and enclose theleg portions144 and thepermanent magnets148 of themagnetic filter assembly138. A similar housing (not shown) may be provided for themagnetic filter assembly140.
The[0068]magnetic filter assemblies138,140 are constructed to fit together as shown in FIG. 17 to createmagnetic filter assembly142. When themagnetic filter assemblies138,140 are interengaged with one another to formassembly142, the magnetic field created by theassembly142 depends on how thecomponent assemblies138,140 are operated. For example, theassemblies138,140 which make up theassembly142 could be operated so that the magnetic fields created by theassemblies138,140 coincide with one another and therefore reinforce one another. For example, it can be understood that themagnetic field assembly140 in FIG. 17 could be energized (by energizing coils153) such that the field created by theassembly140 forms a north pole at the top of the page. Theassembly138 could impose a magnetic field (by energizing the coils150) in which the north pole is at the top of the page or the bottom of the page, depending on the direction of the currents in thecoils150. Thus, it can be appreciated thatcomponent assemblies138,140 ofmagnetic filter assembly142 could be arranged and operated such that the north poles of the twoassemblies138,140 are on the same side of theassembly142, or such that the north poles of the twocomponent assemblies138,140 are on opposite sides of theassembly142.
When the[0069]assembly142 is operated such that the north poles of the twocomponent assemblies138,140 are on the same side of the assembly142 (that is, such that the north poles or magnetic fields of the twocomponent assemblies138,140 are in the same direction or coincide with one another), this arrangement would yield a magnetic field between the layers of plasma similar to that shown in FIG. 14. Alternatively, when theassembly142 is operated such that the north poles of the twocomponent assemblies138,140 are on opposite sides of the assembly142 (or are anti-directional with one another, i.e. the fields are in opposing directions to one another), this arrangement would yield a magnetic field between the layers of plasma similar to that shown in FIG. 12.
In the instance in which the respective magnet fields of the two component[0070]magnetic assemblies138,140 of theassembly142 are in the same direction, a magnet wall type of filter is formed. In the instance in which the respective magnet fields of the two componentmagnetic assemblies138,140 of theassembly142 are anti-directional (that is, so that the respective fields of theassemblies138,140 are in opposing directions to one another), a magnetic cusp filter is created.
The mechanism by which the EEDF (electron energy distribution function) of the[0071]process plasma24 is controlled is considered below. In this example, a magnetic cusp filter is used to control the EEDF. The leakage width for the collision-less fast electrons has been found to be approximately twice their gyroradius, 2 rp, where the gyroradius, rp, is given by the following relationship:
rp=(2eVp/m)1/2/(eB/mc) (Equation 3)
where the primary electron energy is eVp, e is the fundamental charge, m is the electron mass, c is the speed of light in a vacuum and B is the magnetic field strength. The leakage width of the primary electrons is inversely proportional to the magnetic field strength B.[0072]
The leakage rate for the cooled plasma is dominated by the leakage of the ions. Electron leakage is strongly influenced by the ambipolar electric field. In general, the electrons that leak through the magnetic filter are the relatively slow electrons. These relatively slow electrons are further cooled by collisions with the neutral gas, as in the after-glow discharge.[0073]
The apparatuses and methods of the present invention have many advantages. It can be understood from an examination of FIG. 2, for example, that if the[0074]electrode30, themagnetic filter assembly68, the processgas injection device60 and the supporting structures associated therewith were to be removed from theapparatus12, the remaining portions of the apparatus12 (that is, thereaction chamber14, thechuck electrode46, and the transformer coupled plasma source spiral member26) comprise a conventional transformer coupled plasma (TCP) reactor. Conventional TCP reactors have advantages and disadvantages, however, when used for etching or deposition. TCP reactors can be operated to generate high density plasmas, for example, which generally provide a good etch rate, which is desirable, but plasmas generated by conventional TCP reactors have high electron temperatures (typically in the 3-4 eV range), and do not provide the ability to control the electron energy distribution function (EEDF) of the plasma. As a result, these reactors produce high dissociation rates. Generally, the dissociation of a plasma can be related to the density of the plasma, the residence time of a gas atom or molecule moving around in the plasma and to the EEDF of the plasma.
It can be appreciated that although the examples and illustrative embodiments herein use a TCP reactor, this is done to facilitate describing the invention and is not intended to limit the scope of the invention. Other methods and apparatuses may be used to generate the source plasma. For example, the[0075]source plasma20 may in some embodiments of the invention be generated by inductive coupling or, as another alternative, by capacitive coupling.
High dissociation rates are disadvantageous in many plasma process operations. For example, high dissociation rates may be disadvantageous in instances in which the process plasma is used to etch silicon dioxide (which commonly occurs in commercial semiconductor fabrication). Fluorocarbon chemistry is often used to etch the oxide features in a silicon dioxide wafer. These oxide features may include, for example, semiconductor contacts or trenches. A typical combination of gases used to etch a silicon dioxide wafer utilizing a conventional TCP reactor may include a fluorocarbon gas (such as C[0076]4F8, for example), an oxygen containing gas (such as CO, CO2, or O2, for example), and a carrier gas (such as argon, for example). The argon functions to dilute the fluorocarbon-containing and oxygen-containing gases and may also be ionized and used to bombard the surface of the substrate to increase the energy of the etch chemistry.
Often during semiconductor fabrication, for example, the surface of the workpiece has areas of silicon dioxide, photoresist, silicon, silicon nitride and so on. In particular, layers of silicon or silicon nitride can be exposed upon etching through an oxide layer; and, due to process non-uniformities, an over-etch step may be required to complete etching through the oxide layer across the entire substrate. It is often desirable during a semiconductor processing operation to etch the silicon dioxide at a faster rate than the other materials. This requirement is referred to as a requirement for a high degree of etch selectivity for silicon dioxide. Typically, however, inductively coupled plasma sources in general, and TCP reactors in particular, are highly dissociative which leads to the production of relatively high amounts of fluorine radical. When too much fluorine radical is present in the process plasma, the fluorine radicals etch the silicon faster than they etch the silicon dioxide. Thus, a high degree of dissociation in the plasma leads to the formation of a high degree of fluorine radical which leads to a loss of etch selectivity and a consequent degradation in the quality of the semiconductor devices produced.[0077]
Several attempts have been made to reduce the amount of dissociation that occurs in conventional ICP reactors. For example, scavenger materials are sometimes placed in the plasma source. An example of a scavenger material would be silicon which is sometimes provided in the form of a plate. The silicon plate erodes during processing, thereby putting silicon into the plasma chemistry. The silicon reacts with some of the fluorine radical and thereby removes some fluorine radical from the plasma. It may also be desirable to “clad” the surfaces of the chamber with a material or materials suitable for carrying out a specific process. For example, during an oxide etching operation, such materials might include silicon, quartz, and so on. These materials may also be used for other reasons as well.[0078]
The apparatuses and methods of the present invention reduce unwanted dissociation and consequently reduce the amount of fluorine radical in the[0079]process plasma24. This improves etch selectivity. Thus, the plasmapotential controlling electrode36, themagnetic filter assembly68 and the processgas injection device60 can be operated to improve the properties of theprocess plasma24. The ability to control the magnetic field strength of themagnetic field assembly68, for example, provides the operator with the ability to control the energy of the electrons diffusing into theprocess plasma24. This provides the capability of controlling the EEDF of theprocess plasma24. The potentialcontrolling electrode36 provides the ability to control the potential of thesource plasma20. By varying the DC or RF voltage of thepotential controlling electrode36, the ion energy of the ions moving from thesource plasma20 to theprocess plasma24 can be controlled. By controlling the EEDF and the ion energy of theprocess plasma24, the operator can reduce the amount of the dissociation that occurs while keeping the density of theprocess plasma24 relatively high.
More specifically, it can be understood from FIG. 2, for example, that the[0080]electrode36, thefilter assembly68 and the processgas injection device60 essentially divide theinterior area16 of thereaction chamber14 into the tworegions18,22. To process a workpiece (such as a silicon dioxide wafer) utilizing theapparatus12, argon or a similar source gas could be injected in thesource region18 to produce (using, for example, the energized spiral member26) a relatively high density argon plasma. The argon plasma would include positively charged argon ions and relatively high energy electrons. The process gas may be introduced through the processgas injection device60 into theprocess region22 of thechamber14. The process gas may include the fluorocarbon species and perhaps some argon and/or oxygen containing gases. In other words, the processgas injection device60 introduces the gases that will ultimately be dissociated to produce the etch chemistry of the plasma that strikes theworkpiece48. Themagnetic filter assembly68 allows the relatively low energy electrons to enter the process region, but blocks the relatively high energy electrons from entering the process region, thereby confining them to thesource region18. The process gases or species in theprocess region22 interact with these relatively low energy electrons in theprocess region22, but these electrons are not highly dissociative. Thus, if C4F8were introduced into theprocess region22, the energy of the electrons that enter theprocess region22 could be controlled to control the dissociation chemistry of the C4F8to minimize the amount of fluorine radical produced in theprocess plasma24.
The plasma[0081]potential controlling electrode36 controls the potential of thesource plasma20 and cooperates with themagnetic filter assembly68 to control the diffusion of the plasma from thesource region18 into theprocess region22. The presence of a magnetic filter in the center of thechamber14 inhibits the source plasma from diffusing into theprocess region22. Generally, the imposed magnetic field controls the passage of electrons from the source plasma into the process plasma by inhibiting the passage of relatively high energy electrons from the source plasma into the process plasma but allowing the passage of relatively low energy electrons from the source plasma into the process plasma. And generally, thepotential controlling electrode36 helps control the passage of ions into the process plasma by controlling the potential of the source plasma. The plasma density of the source plasma will typically, for example, be much higher then the plasma density of the process plasma. The presence of the potential controlling electrode provides the source plasma with an essentially uniform potential and provides the operator with the ability to adjust the potential of the source plasma.
Because the ions are much more massive then the electrons, they are less affected by the magnetic fields imposed by the magnetic filter assembly. Consequently, if the potential of the source plasma were not controlled to control the movement of the ions from the source plasma into the process plasma, the process plasma could acquire undesirable properties or the[0082]plasma20,24 in thereaction chamber14 could become unstable or turbulent. For example, if the potential of thesource plasma20 were not controlled, initially a relatively high number of positive ions from the source plasma could diffuse into the process plasma while some electrons are inhibited from passing into the process region by the magnetic field. As a result, the process plasma would tend to acquire a positive space charge. This would eventually tend to repel additional positive ions from migrating into theprocess plasma24 which would limit plasma density of theprocess plasma24. The density of theprocess plasma24 can be increased by adjusting the magnetic field to allow migration of electrons from thesource plasma20 into theprocess plasma24. If too many electrons are allowed to pass into theprocess plasma24, however theprocess plasma24 may develop a negative space charge. A negative space potential in theprocess plasma24 may tend to attract positive ions from thesource plasma20 into theprocess plasma24. If the potential of thesource plasma20 were not controlled, these processes (that is the migration of the positive ions and the migration of the negative electrons) could occur in an uncontrollable manner and this may result in turbulence or instability in theplasma20,24 in thechamber14. Thepotential control electrode36 allows the operator (or the controller) to adjust the potential of thesource plasma20 so that charged particles in plasma flow from the source region into the process region in a steady and controlled manner. By controlling the potential of the source plasma and the strength of the magnetic field, the conditions can be adjusted so that ions with high energy and a relatively large current of relatively cold electrons flow into fromsource plasma20 into theprocess plasma24. In general, ions coming to the process region with a controllable directed energy may enhance plasma flow to the wafer and therefore enhance processing efficiency.
It can be understood that the[0083]system10 and the apparatus12 (including the various components thereof shown in the FIGS. 3-17) are example embodiments that are intended to illustrate the principles of the invention, but which are not intended to limit the scope of the invention. Alternative arrangements and additional embodiments are contemplated and within the scope of the invention. For example, although theexample apparatus10 includes three separate components to control the potential of the source plasma, to impose the magnetic fields and to inject the process gas (i.e., the plasmapotential controlling electrode36, themagnetic filter assembly68 and the processgas injection device60, respectively), other arrangements are contemplated. For example, the single or double picket fence embodiments of the magnetic filter assembly could be constructed so that the lower level of the picket fence electrode is constructed to form a gas injection manifold for the injection of the process gas into the process chamber (and thereby perform the function of the process gas injection device60). Additionally, or alternatively, the double or single picket fence could be shaped to spatially adjust the plasma. Additionally, or alternatively, the picket fence could be rotated during a plasma processing operation to make the process plasma more uniform. It is also contemplated to control the electron temperature of the process plasma by introducing a molecular species having a low excitation and ionization potential into the process plasma to absorb electron energy and to radiate energy
It is also contemplated that the plasma[0084]potential controlling electrode36, themagnetic filter assembly68 and the processgas injection device60 could be constructed to be assembled to one another as a unit or module which could then be removably mounted in the reaction chamber of an ICP (or TCP) reactor as a portable module. This modular construction would allow a conventional ICP reactor to be converted for use in performing the methods of the present invention by installing the module within the reaction chamber. For example, such a module could be placed in a TCP in a position to divide the interior of the chamber into a source region and a processing region. A module may include a process gas injection device to inject a gaseous process material into the process region, a potential controlling electrode to control the potential of the source plasma, and a magnetic filter assembly operable to impose a magnetic field generally between the source region and the process region to control the flow of charged particles from the source plasma into the gaseous process material to generate a process plasma in the process region.
The many features and advantages of the present invention are apparent from the detailed specification and thus, it is intended by the appended claims to cover all such features and advantages of the described method which follow in the true spirit and scope of the invention. Further, since numerous modifications and changes will readily occur to those of ordinary skill in the art, it is not desired to limit the invention to the exact construction and operation illustrated and described. Moreover, the method and apparatus of the present invention, like related apparatus and methods used in the semiconductor arts that are complex in nature, are often best practiced by empirically determining the appropriate values of the operating parameters, or by conducting computer simulations to arrive at best design for a given application. Accordingly, all suitable modifications and equivalents should be considered as falling within the spirit and scope of the invention.[0085]